More on why a beanbag chair is a chair.
It's because it has a Family Resemblance to a prototypical chair (Prototypes for Concepts). However, all that means is it has properties that overlap *enough*.
Which properties?
To Lakoff and Johnson, the ones that count are those that capture our typical interactions with chairs.
Things like perceptual properties (how chairs generally look, feel, etc.) can be used, but they don't necessarily capture why chairs *matter*.
* What functions does this object allow you to perform? (sitting) * How do we move when interacting with the object? (what do we do when getting into and out of them, and while we're in them). * What purposes does the object afford (resting, mainly, while eating, relaxing, etc. – all things that would be harder standing up). * What are we using the object for *right now*? (See How Can a Statement Be True.)
Contrast interactional properties with inherent properties. The latter are what's primarily used in dictionary definitions, but the latter is more likely to be what the brain uses.